Everyone shops for the cheapest possible failover box and I get it, money is money. But idk why we do this to ourselves. I just spent 6 hours at 3am recovering from a primary node failure where the "backup" was on the same upstream, same datacenter, basically same everything. Wrote this while waiting for zfs send to finish:
# 3:47 AM, coffee #4
rsync -avz --progress /mnt/tank/backup/ root@deadbox:/mnt/panic/
# ETA: 4 hours 12 minutes
Anyway my point is you are not paying for storage. You are paying for *independence*. Different provider, different network, different power feed. The backup that matters is the one that survives the thing that kills your primary. 💀
Idk why I do this but I ran the numbers last month. My "cheap" backup strategy cost me 8 hours of downtime. My new expensive one would have been 20 minutes. The math is rude.